Countries need “redundancy” and “diversification” in the large language models and other AI tools they use for security and critical systems, Prime Minister Mark Carney said, citing the restrictions the U.S. placed on Claude-maker Anthropic’s Mythos and Fable products as an example of the problem with “overreliance on certain models.” (The Logic)

Talking point: Anthropic said late Friday that the Trump administration, citing national security concerns, had imposed new export controls on the company, preventing it from letting any non-U.S. citizen—including its own staff—from using the two advanced models. While Anthropic had only just launched Fable, businesses across much of the world use its technology. Canadian and European policymakers cited the restrictions as proof of the need for the sovereign AI capabilities they’re looking to spur at home. Toronto-based Cohere, meanwhile, claimed its AI products, which are increasingly open source and can run on clients’ hardware, offer customers more choice and control. Cohere CEO Aidan Gomez said the U.S. move should be a “massive wake-up call.”